Ultrasonic Ice Protection Systems: Analytical and ...
Transcript of Ultrasonic Ice Protection Systems: Analytical and ...
HAL Id: hal-01861799https://hal.archives-ouvertes.fr/hal-01861799
Submitted on 25 Aug 2018
HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.
Ultrasonic Ice Protection Systems: Analytical andNumerical Models for Architecture Tradeoff
Marc Budinger, Valérie Pommier-Budinger, Gael Napias, Arthur Costa daSilva
To cite this version:Marc Budinger, Valérie Pommier-Budinger, Gael Napias, Arthur Costa da Silva. Ultrasonic Ice Pro-tection Systems: Analytical and Numerical Models for Architecture Tradeoff. Journal of Aircraft,American Institute of Aeronautics and Astronautics, 2016, 53 (3), pp.680 - 690. �10.2514/1.C033625�.�hal-01861799�
- 1 -
Ultrasonic ice protection systems: analytical and numerical models for
architecture trade-off
Marc Budinger(1)
, Valérie Pommier-Budinger(2)
, Gael Napias(2)
, Arthur Costa Da Silva(2)
(1) INSA Toulouse, Institut Clément Ader, Toulouse, 31077, France
(2) ISAE SUPAERO, Institut Supérieur de l'Aéronautique et de l'Espace, 31055, France
ABSTRACT
Protection systems against ice conventionally use thermal, pneumatic or electro-thermal solutions. However, they are
characterized by high energy consumption. This article focuses on low-consumption electromechanical deicing solutions
based on piezoelectric transducers. After a review of the state of the art to identify the main features of electromechanical
de-icing devices, piezoelectric transducer-based architectures are studied. Analytical models validated by numerical
simulations allow trend studies to be performed which highlight the resonance modes and the ultrasonic frequency
ranges that lead to low-consumption, compact ultrasonic deicing devices. Finally, de-icing systems widely studied with
bonded ceramics and de-icing systems less usual with Langevin pre-stressed piezoelectric transducers are compared and
a Langevin piezoelectric transducer-based device leading to an interesting compromise is tested.
Keywords: de-icing systems, electromechanical actuator, ultrasonic actuator, piezoelectric transducer, pre-stressed
Langevin actuator
1. INTRODUCTION AND STATE OF THE ART
1.1. Ice protection systems
Ice accretion on aircraft has been a well identified problem since the beginning of the 20th
century. It can lead to
decreased lift, increased drag, reduced thrust reduction, and risk of stalling or even engine failure owing to ice ingestion.
Icing occurs both during flight and on the ground. It has led to many aviation accidents, such as the Air Florida Boeing
B737 (1982), the American Eagle ATR 72 (1997), and the Air France Airbus A330 (2009). To ensure aircraft safety,
regulatory agencies, such as the CAA (Civil Aviation Authority [1]) and the FAA (Federal Aviation Administration [3]),
have established regulations for aircraft anti-icing and de-icing.
- 2 -
Current strategies for anti-icing and de-icing [1] can be chemical, thermal or mechanical - each having different
degrees of efficiency or environmental impact. The chemicals used for de-icing (ethylene, propylene glycol or diethylene
glycol) can lower the freezing point but require large volumes of fluids and induce environmental issues and premature
wear of the treated parts (especially corrosion). Thermal techniques are used for anti-icing and de-icing in flight and on
the ground by liquefaction and vaporization of the ice, but require either a large amount of hot air under pressure to be
bled from the engine or a large amount of energy to be provided by the electric grid for electro-thermal solutions.
Thermal solutions require around 4 kW/m² and the total amount of power required to de-ice a Boeing 787 with an
electrical de-icing system has been estimated at 76 kW [11]. Mechanical de-icing systems are low-energy solutions that
aim to break the accumulated ice by applying a mechanical pulse or vibrations to the structure to be protected. Pneumatic
systems are commonly used for their low cost but have a significant impact on the aerodynamics of the aircraft and
require maintenance. Recent efforts to develop electromechanical systems are justified by the potential of such systems in
terms of weight, durability and energy savings.
1.2. Electromechanical de-icing systems
Electromechanical deicing systems use electromagnetic actuators, piezoelectric actuators or shape memory
alloys. These systems are presented in [1] and [6] and their advantages and drawbacks are analyzed in Table 1.
This article focuses on actuation with piezoelectric technology, and especially on resonant piezoelectric
technologies, which have a better power/mass ratio than static solutions. The deeper insight into piezoelectric
de-icing systems presented in Table 2 allows a more detailed analysis to be carried out according to the frequency
range of the modes activated by the piezoelectric actuators and to the type of piezoelectric actuators.
Ramanathan et al. [7] proposed the use of ultrasonic shear waves at very high frequency (1 MHz). They performed
experiments with piezoelectric patches bonded to an isotropic plate with a layer of ice. The results indicate that the
actuators were able to de-ice the aluminum plate by melting the ice at the interface. Kalkowski et al. [8] analyzed the
frequency range for which wave-based technologies efficiently promote the delamination of ice with minimum power
requirements.
Venna et al. [9][10][11][12] used piezoelectric ceramics bonded onto plates and on the inner flat surface of a leading
edge structure to excite low frequency modes to delaminate ice (below 1000 Hz). They used analytical and numerical
- 3 -
models to identify the first modes for which the shear stress produced in the ice was greater than the shear stress that
would theoretically lead to delamination. The average de-icing time varied between 46 s and 280 s and increased as the
icing temperature decreased. Palacios [13] analyzed this result and found that the de-icing time seemed to show that the
de-icing was more probably caused by thermal effects than by shear stress. S. Struggl et al. [14] conducted the same kind
of analysis and experiments with piezoelectric ceramics bonded to a plate and on a leading edge structure to excite low
frequency modes (below 500 Hz). They also performed tests in an icing research tunnel and the de-icing was successful at
a frequency of 307 Hz.
Seppings [15] used a stack of thin piezo-electric discs held in compression by a bolt running through the center of the
stack and showed that the pre-stressed actuator driven at 20 kHz was more efficient than piezoelectric patches.
Palacios initiated many studies on de-icing systems [16][17][18][19][20][21] and tested several technologies. For
piezoelectric de-icing systems, he used piezoelectric patches to generate ultrasonic shear stress at high frequency (around
a few tens of kHz). He performed tests on plates [16][17] and on leading edges [18]. At such frequencies, the
delamination of the ice was instantaneous. He also tested an original design of a shear tube actuator driven at 300 V and
436 Hz [19]. In [20], experiments showed two main failures related to the bonding of the actuators: fracture of the
piezoelectric ceramic at the solder joint location and delamination between the ceramic and the host. To tackle this
problem, Palacios proposed optimizing the bonding of the supply wire on the ceramic and the bonding between the
piezoelectric actuators and the substrate to avoid concentrations of stresses [21]. In [22], he also investigated the effect of
hydrophobic coating combined with an ultrasonic deicing system and showed that the ice adhesion depended on the
substrate roughness.
More recently, Strobl [23] used multilayer piezoelectric patches at frequencies around 4 kHz and icephobic coating to
delaminate ice instantaneously on polished surfaces with a low supply voltage.
- 4 -
Technological solutions Main advantages Main drawbacks
Electromagnetic
Contactless possible, high
displacements possible
Low force density, electromagnetic pollution,
size of feeding electronics
Piezoelectric (PZT)
Force generation (good force/mass
ratio) even at high frequency,
energy consumption
Brittleness, small displacements
Shape Memory Alloy
(SMA)
Force generation in static Low dynamics, response time, energy
consumption
Table 1 – Comparison of electromechanical deicing systems
Comparison
criteria
Technological solutions Previous works
Main
advantages
Main drawbacks
Frequency
range
Static, Low Frequency
(Hz)
[9][10][11][12] [14]
Power
supply
Force of activation
Vibration (kHz)
[16][17] [18][19][20][21]
[22][23]
Fatigue
Waves (MHz) [7][8]
Energy
consumption
Technology
Bonded Piezoelectric
ceramics (PZT or
multilayers)
[9][10][11][12][15]
[16][17][18]
[20][21][21][22][14][23]
Easy to
implement
Brittle
Not easily
adaptable to
curved surfaces
Pre-stressed piezoelectric
transducer
[15]
Less brittle,
stress-
resistant
Frequency linked
to the size of the
transducer
Table 2 – Comparison of piezoelectric de-icing systems
- 5 -
1.3. Review of previous works, proposed architecture and objectives of present work
According to the previous analyses, the main features of piezoelectric de-icing devices are as follows:
- As regards the frequency range, previous studies on piezoelectric de-icing devices have shown that both low
frequencies and ultrasonic high frequencies can be excited to generate shear stress and to break ice. However, no
studies have been carried out to determine the best frequency range for which structural resonances maximize the
shear stress in the ice while minimizing the stress in the substrate and in the actuators. Consequently, a specific study
to determine the optimized frequency range is required and is proposed in this article.
- One other issue is the nature of the modes to be excited: both flexural and extensional modes can produce shear
stress in a structure. Thus, a specific study to determine the kind of modes that produce maximal shear stress will be
useful for the design of the de-icing system architecture.
- Regarding the actuation technology, this article will compare non-pre-stressed piezoelectric transducers (bonded
ceramics) and pre-stressed piezoelectric transducers (Langevin transducer) in terms of force-to-density ratio,
robustness, ease of integration in a curved leading edge and power consumption. Non-pre-stressed piezoelectric
transducers have been widely tested and examples off architectures can be found in [9][15][16][18][23]. Figure 1
shows two different architectures with a pre-stressed piezoelectric actuator, one exciting flexural modes, the other
exciting extensional modes. Both architectures lead to shear stress at the ice/substrate interface.
Figure 1 - Architecture for a de-icing system with a pre-stressed piezoelectric transducer: (a) Excitation of
flexural modes, (b) Excitation of extensional modes
The objectives of this paper are thus:
- to develop analytical models validated by numerical simulations and complemented by measurements in order to
perform trend studies highlighting the nature of the resonance modes and the frequency ranges leading to low-
consumption, compact piezoelectric de-icing devices.
Ice Aluminum substrate
Ice
Aluminum substrate
(a)
(b)
- 6 -
- to compare piezoelectric transducer-based architectures enabling an interesting compromise to be reached to
generate shear stress at the ice/substrate interface and to promote ice delamination and cracking without damaging the
actuator or the substrate.
2. FEASIBILITY OF PIEZOELECTRIC DE-ICING SYSTEMS BASED ON STRESS ANALYSIS
2.1. Study case
In this article, the study case is a simple plane rectangular plate of dimensions 290 × 200 × 1.5 mm3 that is covered
by a 2mm thick layer of ice. The panels of an aircraft airframe are usually supported by ribs and stiffeners and the
boundary conditions of such panels are close to clamped on all sides. However, the plates will be studied in the pinned
boundary condition in this section because this condition enables simple analytical expressions to be formulated to
estimate the stresses produced at the ice/substrate interface and because, at the high frequencies used in the study, the
deformations in the middle of the plate are similar for pinned and clamped boundary conditions (as illustrated in Figure
2).
Figure 2 - Comparison of two boundary conditions (clamped versus pinned) for the study of a plate at high
frequencies: displacements in the middle of the plate are sinusoidal in both cases
The mechanical properties of the ice considered for calculation were chosen so as to correspond to glaze ice and were
selected among the values presented in [24]. The ice and aluminum mechanical properties used for the analyses
performed in this article are shown in Table 3.
f = 12 254 Hz
Clamped boundary case
145mm middle of the plate
f = 11 434 Hz
145mm middle of
the plate
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 7 -
Material Aluminum Ice
𝐸 (𝐺𝑃𝑎) 70 9.7
𝐺 (𝐺𝑃𝑎) 26 3.7
𝜈 (−) 0.33 0.30
𝜌 (𝑘𝑔/𝑚3) 2770 880
Table 3 –Mechanical properties of the study case materials
2.2. Stress generation
In order to choose the vibration modes that are the most suitable for de-icing by generation of stresses in a plate, four
types of “failure” modes of the ice/substrate interface were considered: failure owing to excessive tensile stress; failure
owing to excessive in-plane shear stress; failure owing to excessive out-of-plane shear stress and failure owing to
excessive out-of-plane tensile stress. Figure 3 shows a simplified scheme of the mechanisms and the interface failure
expected in each case. The values of ice strength are discussed in [25],[26],[27]. The parametric studies carried out by
Scavuzzo et al. [25] and by Jellinek [26] and the experiments performed by Laforte et al. [27] indicate that the adhesion
strength of ice depends mainly on the roughness of the accretion surface. Scavuzzos’s study also suggests that the
adhesion resistance to an out-of-plane stress (shear or tensile) depends, but to a lesser degree, on droplet momentum and
surface temperature. Laforte’s work shows that the adhesion resistance to an in-plane deformation (distortion or
elongation) varies with the ice thickness and Loughborough observed dependence of the strength on the nature of the
substrate [28]. Finally, Table 4 exposes the values of the properties presented in the references discussed. In this study,
the targeted application is the de-icing of aircraft flight control surfaces. For this kind of application, ice delamination by
shear is more efficient than breaking by tensile stress. This is why we focus on out-of-plane shear stress generation. For
the assessment of the different de-icing system architectures, we will compute the out-of-plane shear stress and ice will be
assumed to be de-bonded when the out-of-plane shear stress exceeds the critical adhesive strength of ice (considered to be
1 MPa for numerical applications of this paper). Moreover, to assess de-icing systems, the stress induced in the substrate
and in the piezoelectric ceramics will also be computed in order to quantify the risk of damage to the ceramics.
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 8 -
Figure 3 - Modes of failure of the ice-aluminum interface. a) Failure due to excessive in-plane tensile stress. b)
Failure due to excessive in-plane shear stress. c) Failure due to out-of-plane shear stress. d) Failure due to
excessive out-of-plane tensile stress
Ice Thickness (mm) 2 5 10
De-icing Tensile Strain ϵx,crit(μm/m) 500 420 280
De-icing In-Plane Shear Strain ϵxy,crit(μm/m) 700 350 212
De-icing Out-of-Plane Shear Stress τzx,crit(MPa) 1.10 – 0.55
De-icing Out-of-Plane Tensile Stress σz,crit(MPa) 1.20 – 0.95
Ice Tensile Strength (MPa) 3.1 – 0.7
Ice Compressive Strength (MPa) 5-25
Ice Shear Strength (MPa) 0.7
Table 4 –Adhesion properties of impact ice to aluminum surface with matte finish and its strength properties
([25],[26],[27])
2.3. Comparison of resonance modes on 1D models
The plate resonance modes used in the literature (section 1.2) are essentially in-plane extensional modes and out-of-
plane flexural modes. They generate stress corresponding to failure modes a) and c) in Figure 3. However, piezoelectric
actuators have limited displacement capacity. Consequently the comparison of these modes is made here in terms of
stresses generated for a given displacement. To simplify the analysis, several assumptions are used: support and ice are
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 9 -
considered as a thin multilayer beam (1D model), mode shapes are assumed to be identical to those of a uniform beam
and boundary conditions are simply supported as introduced in section 2.1. Figure 4 shows the beam under study where:
x is the transverse position along the beam of length a ;
n is the number of anti-nodes for the mode considered;
is the pulsation of the mode considered;
halu, hice, hn are respectively the thickness of the aluminum beam, the thickness of the ice beam and the position of the
neutral line for the flexural mode,
U(x) and W(x) are respectively the in-plane displacements (for extensional modes) and out-of-plane displacements
(for flexural modes);
calu , cice , alu and ice are respectively the Young modulus and the density for the aluminum beam and the ice.
a)
b)
Figure 4 –Element of the beam under study
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 10 -
For flexural modes, the position of the neutral line hncan be obtained by assuming that the tensile force in a section is
zero. This results in:
ℎ𝑛 =1
2
𝑐𝑎𝑙𝑢ℎ𝑎𝑙𝑢2 − 𝑐𝑖𝑐𝑒ℎ𝑖𝑐𝑒
2
𝑐𝑎𝑙𝑢ℎ𝑎𝑙𝑢 + 𝑐𝑖𝑐𝑒ℎ𝑖𝑐𝑒
(1)
Table 5 synthetizes the analytical equations derived from [29] to compute strains and stresses in the ice, in the
aluminum substrate and at the ice-substrate interface. The equations of the peak out-of-plane shear stress were obtained
by isolating an element of ice of thickness dx subjected to elastic forces, inertial forces and shear forces at the ice-
substrate interface. The resonance frequencies were obtained by the Rayleigh method [30] using the expression of the
strains to estimate the kinetic and potential energies.
Extensional modes Flexural modes
Displacement 𝑢(𝑥, 𝑡) = 𝑈(𝑥) 𝑠𝑖𝑛 𝜔𝑡 = 𝑈0 𝑠𝑖𝑛𝑛𝜋𝑥
𝑎𝑠𝑖𝑛 𝜔𝑡 𝑤(𝑥, 𝑡) = 𝑊(𝑥)𝑠𝑖𝑛 𝜔𝑡 = 𝑊0 𝑠𝑖𝑛
𝑛𝜋𝑥
𝑎𝑠𝑖𝑛 𝜔𝑡
𝑢(𝑥, 𝑡) = −𝑧𝜕𝑤
𝜕𝑥= 𝑈(𝑥)𝑠𝑖𝑛 𝜔𝑡
Peak tensile strain 𝜀𝑥 =
𝜕𝑈(𝑥)
𝜕𝑥= 𝑈0
𝑛𝜋
𝑎𝑐𝑜𝑠
𝑛𝜋𝑥
𝑎 𝜀𝑥 =
𝜕𝑈(𝑥)
𝜕𝑥= 𝑊0𝑧 (
𝑛𝜋
𝑎)
2
𝑠𝑖𝑛𝑛𝜋𝑥
𝑎
Peak ice tensile
stress in ice
𝜎𝑥 = 𝑐𝑖𝑐𝑒𝜀𝑥 = 𝑐𝑖𝑐𝑒𝑈0
𝑛𝜋
𝑎𝑐𝑜𝑠
𝑛𝜋𝑥
𝑎 𝜎𝑥 = 𝑐𝑖𝑐𝑒(ℎ𝑖𝑐𝑒 + ℎ𝑛) (
𝑛𝜋
𝑎)
2
𝑊0 𝑠𝑖𝑛𝑛𝜋𝑥
𝑎
Peak tensile stress
in aluminum
𝜎𝑥 = 𝑐𝑎𝑙𝑢𝜀𝑥 = 𝑐𝑖𝑐𝑒𝑈0
𝑛𝜋
𝑎𝑐𝑜𝑠
𝑛𝜋𝑥
𝑎 𝜎𝑥 = 𝑐𝑎𝑙𝑢(ℎ𝑎𝑙𝑢 − ℎ𝑛) (
𝑛𝜋
𝑎)
2
𝑊0 𝑠𝑖𝑛𝑛𝜋𝑥
𝑎
Peak out-of- plane
shear stress at the
ice/aluminum
interface
𝜏𝑥𝑧 = (𝜌𝑖𝑐𝑒𝜔2 − 𝑐𝑖𝑐𝑒 (𝑛𝜋
𝑎)
2
) ℎ𝑖𝑐𝑒𝑈0 𝑠𝑖𝑛𝑛𝜋𝑥
𝑎 𝜏𝑥𝑧 = 𝑐𝑎𝑙𝑢 (
𝑛𝜋
𝑎)
3 (ℎ𝑎𝑙𝑢 − ℎ𝑛)2 − ℎ𝑛2
2𝑊0 𝑐𝑜𝑠
𝑛𝜋𝑥
𝑎
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 11 -
Angular frequency
𝜔 = 𝜋𝑛
𝑎√
𝑐𝑖𝑐𝑒ℎ𝑖𝑐𝑒 + 𝑐𝑎𝑙𝑢ℎ𝑎𝑙𝑢
𝜌𝑖𝑐𝑒ℎ𝑖𝑐𝑒 + 𝜌𝑎𝑙𝑢ℎ𝑎𝑙𝑢
𝜔 = 𝜋2 (
𝑛
𝑎)
2
√𝐸𝐼
𝜌𝑖𝑐𝑒ℎ𝑖𝑐𝑒+𝜌𝑎𝑙𝑢ℎ𝑎𝑙𝑢 with
𝐸𝐼 =𝑐𝑎𝑙𝑢
3(ℎ𝑛
3 + (ℎ𝑎𝑙𝑢 − ℎ𝑛)3)
+𝐶𝑖𝑐𝑒
3((ℎ𝑛 + ℎ𝑖𝑐𝑒)3 − ℎ𝑛
3)
Table 5 – 1D modes equations
These analytical equations show that:
- for extensional modes, the maximum tensile stress in the substrate or the ice is located on the displacement nodes and
the maximum out-of-plane shear stress at the ice-substrate interface is located on the displacement antinodes,
- for flexural modes, the maximum out-of-plane shear stress is located on the displacement nodes.
Figure 5 also compares the extensional and flexural modes, particularly their ability to generate shear stress at the ice-
substrate interface. It also gives the tensile stress in the ice and in the aluminum substrate. It shows that, for a given
frequency and for a given displacement, the shear stress level at the ice-substrate interface is smaller for extensional than
for flexural modes, while the tensile stress in the aluminum substrate is almost the same. Thus, to generate ice
delamination while minimizing the displacement of the piezoelectric actuators, it is more interesting to excite flexural
modes. This is done for example by architecture a of Figure 1. Another conclusion that can be drawn from Figure 5 is that
it is more favorable to work at high frequencies: the higher the frequency, the higher the shear stress. However, the
frequency of use may be limited by other criteria such as power supply issues.
Note: the results of Figure 5 were obtained for a 2 mm thick ice layer. The expression of out-of-plane shear stress xz
shows that this stress would be maximum for a null position of the neutral line (hn=0), which implies an optimal ice
thickness hice:
ℎ𝑖𝑐𝑒 = ℎ𝑎𝑙𝑢√𝑐𝑎𝑙𝑢
𝑐𝑖𝑐𝑒
(2)
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 12 -
Relation (2) gives an optimal ice thickness of 4.5 mm for a 1.5 mm thick aluminum substrate. However, for future
calculations we chose a 2 mm thick ice layer, less attractive in terms of stress generation, but more realistic for the
targeted application (de-icing of aircraft flight control surfaces).
Figure 5 – Comparison of extensional and flexural modes
2.4. Shear stress estimation on 2D models for flexural modes
The previous section gave 1D models of beam type structures. In this section, stresses are expressed for 2D plate type
geometries. Following the conclusions of the study for 1D models, stresses will be computed only for flexural modes. For
this type of movement, each point of the plate is considered to have a vertical displacement ww in its x and y directions
such that 𝑤 =w = w(x, y, t). We assume that the displacement w(x, y, t) can be approximated by the analytical solution
for flexural modes of a homogeneous plate [30]. Table 6 synthetizes the analytical equations to compute the strains, the
shear stress at the ice-substrate interface and the tensile stresses in the aluminum substrate and in the ice. n is the number
of anti-nodes on the length for the considered mode and m the number of anti-nodes on the width.
10-2
10-1
100
101
102
103
10-5
10-4
10-3
10-2
10-1
100
101
102
Frequency (kHz)
Sh
ea
r (.
..)
an
d te
nsile
(_
__
) S
tre
ss (
MP
a/µ
m)
Stress vs frequency
Shear at ice/substrate interface (flexural mode)
Tensile in aluminum substrate (flexural mode)
Tensile in ice (flexural mode)
Shear at ice/substrate interface (extensional mode)
Tensile in aluminum substrate (extensional mode)
Tensile in ice (extensional mode)
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 13 -
Flexural modes
Displacements 𝑤(𝑥, 𝑦, 𝑡) = 𝑊0 sin𝑛𝜋𝑥
𝑎sin
𝑚𝜋𝑦
𝑏sin 𝜔𝑡
𝑢(𝑥, 𝑦, 𝑡) = −𝑧𝜕𝑤
𝜕𝑥
𝑣(𝑥, 𝑦, 𝑡) = −𝑧𝜕𝑤
𝜕𝑦
Strains
Peak tensile strain
(Sxx)
𝜀𝑥 = 𝑊0𝑧 (𝑛𝜋
𝑎)
2
sin𝑛𝜋𝑥
𝑎sin
𝑚𝜋𝑦
𝑏
𝜀𝑦 = 𝑊0𝑧 (𝑚𝜋
𝑏)
2
sin𝑛𝜋𝑥
𝑎sin
𝑚𝜋𝑦
𝑏
Peak in-plane
shear strain (Sxy)
𝛾𝑥𝑦 =1
2(
𝜕𝑢
𝜕𝑦+
𝜕𝑣
𝜕𝑥) = −𝑧
𝜕2𝑤
𝜕𝑥𝜕𝑦= −𝑧𝑊0
𝑛𝜋
𝑎
𝑚𝜋
𝑏cos
𝑛𝜋𝑥
𝑎cos
𝑚𝜋𝑦
𝑏
Stresses
Peak ice tensile
stress (Txx)
𝜎𝑥 =𝑐𝑖𝑐𝑒
1 − 𝜈𝑖𝑐𝑒2
(ℎ𝑖𝑐𝑒 + ℎ𝑛) ((𝑛𝜋
𝑎)
2
+ 𝜈 (𝑚𝜋
𝑏)
2
) 𝑊0 sin𝑛𝜋𝑥
𝑎sin
𝑚𝜋𝑦
𝑏
Peak aluminum
tensile stress (Txx)
𝜎𝑥 =−𝑐𝑎𝑙𝑢
1 − 𝜈𝑎𝑙𝑢2
(ℎ𝑎𝑙𝑢 − ℎ𝑛) ((𝑛𝜋
𝑎)
2
+ 𝜈 (𝑚𝜋
𝑏)
2
) 𝑊0 sin𝑛𝜋𝑥
𝑎sin
𝑚𝜋𝑦
𝑏
Peak in-plane
shear stress (Txy)
𝜏𝑥𝑦 = −(ℎ𝑖𝑐𝑒 + ℎ𝑛)𝑐𝑖𝑐𝑒
1 + 𝜈𝑖𝑐𝑒
𝑊0
𝑛𝜋
𝑎
𝑚𝜋
𝑏cos
𝑛𝜋𝑥
𝑎cos
𝑚𝜋𝑦
𝑏
Peak out-of- plane
shear stress (Txz)
𝜏𝑥𝑧 =𝑐
1 − 𝜈2((
𝑛𝜋
𝑎)
2
+ (2 − 𝜈) (𝑚𝜋
𝑏)
2
)𝑛𝜋
𝑎
(ℎ𝑎𝑙𝑢 − ℎ𝑛)2 − ℎ𝑛2
2𝑊0 𝑐𝑜𝑠
𝑛𝜋𝑥
𝑎𝑠𝑖𝑛
𝑚𝜋𝑦
𝑏
Table 6 –2D flexural modes equations
These equations are validated by a Finite Element analysis performed for the geometry of the study case (halu = 1.5
mm, calu = 70 GPa, hice = 2 mm, cice = 9.7 GPa, hn = 0.48 mm, a=290 mm, b = 200 mm). Table 7 shows the mode for n=13
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 14 -
and m=5 of frequencies around 15 kHz, which is an intermediate frequency among the frequencies usually used for tests
with piezoelectric ceramics bonded on substrate. The comparison of analytical and numerical results shows a difference
of less than 8% and validates the analytical models developed.
Flexural mode
n=13
m=5
Analytical results Numerical results
Resonance frequency 15554 Hz 14732 Hz
Maximum tensile stress in the ice layer 0.57 MPa/µm 0.52 MPa/µm
Maximum shear stress at the ice/substrate
interface
0.134 MPa/ µm 0.144 MPa/µm
Table 7 – Comparison of analytical equations and Finite Element Analysis
2.5. Frequency range for piezoelectric de-icing systems
The equations of Table 6 will be used to study the feasibility of de-icing systems with piezoelectric actuators and to
highlight the frequency ranges leading to efficient piezoelectric deicing devices. The study will be performed for the
study case of section 2.1, with a number of anti-nodes in the length varying between 1 and 25 and a number of anti-nodes
in the width varying between 1 and 15.
Figure 6 and Figure 7 show that the shear stress per µm of displacement at the ice/substrate interface and the tensile
stress per µm in aluminum or ice increase with the number of anti-nodes and with frequency. One requirement for de-
icing systems is to remove ice without stressing the structure on which it is deposited. This means that de-icing systems
must maximize stress at the ice/substrate interface while minimizing stress in the structure. The ratio of tensile stress in
aluminum to shear stress at the ice/substrate interface represented in Figure 8 allows the frequency range to be found for
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 15 -
which the de-icing systems are the most efficient in meeting this requirement and shows that ultrasonic frequencies,
higher than 20 kHz, are more favorable.
Figure 6 - Flexural resonance frequencies and stresses for the study case versus the number of anti-nodes
Figure 7 - Shear stress per µm according to frequency for flexural modes
Number m of anti-node (200 mm)
Num
ber
n o
f anti-n
ode (
290 m
m)
Frequency (kHz)
5 10 15
5
10
15
20
25
10
20
30
40
50
60
70
Number m of anti-node (200 mm)
Num
ber
n o
f anti-n
ode (
290 m
m)
Shear stress (MPa/µm)
5 10 15
5
10
15
20
25
0.2
0.4
0.6
0.8
1
1.2
Number m of anti-node (200 mm)
Num
ber
n o
f anti-n
ode (
290 m
m)
Tensile stress (MPa/µm) in ice
5 10 15
5
10
15
20
25
0.5
1
1.5
2
Number m of anti-node (200 mm)
Num
ber
n o
f anti-n
ode (
290 m
m)
Tensile stress (MPa/µm) in aluminium
5 10 15
5
10
15
20
25
1
2
3
4
5
6
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 16 -
Figure 8 - Aluminum stress / shear stress according to frequency for flexural modes
3. EVALUATION OF DIFFERENT ARCHITECTURES OF PIEZOLECTRIC DE-ICING SYSTEMS
The previous section highlighted the type of modes (flexural modes) and the frequency range (ultrasonic) of the
resonant modes to be excited to lead to ice delamination. This section aims to assess the type of piezoelectric actuators to
be used. Two kinds of actuators are compared: patch type actuators, directly bonded on to the structure to be activated,
and Langevin pre-stressed actuators in the configuration Figure 1a.
3.1. Methodology for evaluating piezoelectric deicing systems
The proposed methodology for evaluating the 2 different architectures of piezoelectric de-icing systems is divided into
2 main phases:
1. Computation of a reduced model (analytically or numerically) of the chosen architecture of the de-icing system
connected to the surface with the layer of ice. Computation of the maximal tensile stress per µm within the PZT
ceramics and of the shear stress per µm at the ice/substrate interface. All these results are computed for the
resonance mode for which the coupling between the piezoelectric actuator and the structure is the best, i.e. for
which the required voltage will be the lowest.
2. Computation of the displacement required to generate the minimal stress value and of the voltage that leads to the
required displacement. As the piezoelectric de-icing systems are resonant systems, this result depends strongly on
the damping of the structure with the ice.
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 17 -
3.1.1. Reduced model of systems with piezoelectric actuators
This section details the computations of the two design methodology phases. These details are extracted from [34].
The reduced model of a structure with piezoelectric actuators can be made for one mode with a mechanical equation and
an electrical equation:
VCNqq
NVKqqDqM
oC
s
(3)
where q is the modal displacement, M the modal mass, K the modal stiffness, qc the electrical charge, V the voltage, Ds
the modal damping, the modal electromechanical coupling factor, and Co the modal turned-off capacity.
This model can be computed analytically for simple geometries or with multiphysics Finite Element software (such as
COMSOL©, ANSYS© or ABAQUS©) that allows calculations with piezoelectric elements. If the computations are
carried out with short-circuited piezoelectric patches, V=0 and the equations become:
θqq
NVKqqDqM
C
s
(4)
which means that
𝑁 =𝑞𝑐
𝑞
(5)
and for the resonance:
𝑁 =𝑞𝑐
𝑞
(6)
For structures with low damping,
𝑄𝑚 ≈𝐾
𝜔𝐷𝑠
(7)
Consequently, if the displacement required to de-bond ice is known for a resonance mode, the voltage that generates this
displacement is given by:
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 18 -
𝑄𝑚 ≈𝐾
𝜔𝐷𝑠
(8)
3.1.2. Design drivers of piezoelectric actuators
The electromechanical coupling factor N describes the ability of the piezoelectric actuators to generate a force on the
degree of freedom q for a given resonance mode. Moreover, stresses are proportional to displacements of the degrees of
freedom. We chose to compute the electromechanical coupling factor for the point characterized by the largest
displacement (in-plane or out-of-plane depending on excited modes). The electromechanical coupling factor may be
influenced by: the mode of the piezoelectric ceramic of the transducer (mode 31 for the bonded patch and mode 33 for the
Langevin transducer), the dimensions and positioning of the ceramic relative to the nodes and antinodes, and the presence
of adhesive.
A piezoelectric transducer may have several operational limits [35]: maximum voltage to avoid depolarization by
application of an excessive electric field (400 V/mm of ceramic thickness), maximum stress in the ceramics to avoid
failure (about 24 MPa in extension for dynamic applications and for non-pre-stressed ceramics) and maximum
temperature to avoid reaching the Curie temperature, which leads to depolarization. The thermal limit will not be
considered here because of the very low ambient temperatures.
As shown in Equation 30, the displacement at resonance depends on the damping of the structure, characterized by a
mechanical quality coefficient. This coefficient is highly dependent on the boundary conditions and on the complexity of
the assembly. To assess the different architectures, we assume a value of Qm equal to 100 here, which corresponds to a
damping coefficient of 0.5%.
3.2. Piezoelectric de-icing system with bonded ceramics
The first configurations studied here are inspired directly from the work of Palacios ([17][18][20][21]) because he
experimentally proved their ability to provide instantaneous de-icing even for untreated surfaces (without polishing,
without hydrophobic or icephobic coating). The actuator configurations consist of one or more PZT disks glued onto the
surface to be de-iced. Palacios explained that ice delamination occurs for frequencies near the first radial extensional
mode of the piezoelectric disk.
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 19 -
Two simple configurations will be processed in this section:
Free axisymmetric disks described in [36] (Figure 9a). This study, which allows simple 2D Finite Element analyses
because of the symmetry, will help the phenomena involved to be understood on a 2D model and allows the
parameters that are negligible for future simulations in 3D (e.g. glue thickness) to be estimated, thus avoiding fine
meshes of very thin layers and heavy 3D models.
Clamped plates with the dimensions given in section 2.1 (Figure 9b). This geometry will allow a comparison with
the architecture based on a Langevin transducer, which will be studied in the next section. A volume of ceramic
similar to that of the Langevin transducer is chosen in order not to bias the comparison.
Figure 9 - Studied configurations with bonded ceramics (free disk and clamped plate)
We start by the study of the free axisymmetric disks. Figure 10 describes the geometry studied by Soltis under the
direction of Palacios [36]. We performed studies step by step (PZT ceramic only, aluminum plate only, PZT/aluminum
plate, PZT/aluminum plate /ice) to understand the principle of ice delamination.
Figure 10 - Axisymmetric configuration with bonded ceramics (disk)
(a) (b)
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 20 -
The study of the piezoelectric ceramic alone shows that the extensional radial frequency of the PZT disk is around
29.9 kHz. As Palacios recommends studying the modes around the extensional radial mode of the PZT disk, extensional
and flexural modes around this frequency are preferentially studied.
We first look at the modal shape of these 2 modes. Figure 11(a)(b)(c) shows the modal shapes of the aluminum plate
alone (case (a)), the aluminum plate with the piezoelectric disk (case (b)) and the aluminum plate with the piezoelectric
disk and ice (case (c)). These modal shapes are represented for the extensional radial frequency close to that of the PZT
disk. With a PZT disk bonded on the aluminum plate (with or without ice), the configuration is not symmetric about one
plane normal to the axisymmetric axis and the radial mode thus generates both radial and axial displacements. Figure
11(d) shows the modal shape of the flexural frequency close the extensional radial frequency of the PZT disk for the case
of aluminum plate with piezoelectric disk and ice. The axial displacements generated by the flexural mode are phased-
shifted by a quarter of a wavelength compared to the axial displacements generated by the extensional mode. Such axial
displacements generated by the flexural and extensional modes and phased-shifted explain why this configuration excited
by a sweep in frequency is favorable to de-ice the entire surface: nodes of the axial displacements for which shear stress
are important are spread over the entire surface.
Then Table 8 gives (i) the parameters of the reduced model for the flexural and extensional modes, including the
electromechanical force factor, and (ii) the computation results of the shear stresses required to de-bond the ice. The
extensional mode has a high electromechanical force factor; the transducer is well-coupled with the structure. For a
quality factor of 100, the required displacement to de-bond the ice is 9.22µm, which induces a voltage of 124 V - which is
reasonable - and a tensile stress in the aluminum plate of nearly 42 MPa – which is quite high. For the flexural mode, the
electromechanical force factor is much smaller but, as the stiffness is also smaller, the required displacement to de-bond
the ice is 7.72µm and the required voltage is 88 V. The tensile stress in the aluminum plate is also around 42 MPa, similar
to the extensional mode.
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 21 -
(a) Extensional mode for Alu (28 kHz)
(b) Extensional mode for Alu+PZT (28.5 kHz)
(c) Extensional mode for Alu+PZT+Ice (26.2 kHz)
(d) Flexural mode for Alu+PZT+Ice (27.7 kHz)
Figure 11 – Axial and radial displacements for axisymmetric configuration
-1,0
-0,5
0,0
0,5
1,0
1,5
0 0,05 0,1 0,15 0,2
No
rmal
ize
d d
isp
lace
me
nt
Radial position (m)
Axial
Radial
-1,0
-0,5
0,0
0,5
1,0
1,5
0 0,05 0,1 0,15 0,2
No
rmal
ize
d d
isp
lace
me
nt
Radial position (m)
Axial
Radial
-1,0
-0,5
0,0
0,5
1,0
1,5
0 0,05 0,1 0,15 0,2
No
rmal
ize
d d
isp
lace
me
nt
Radial position (m)
Axial
Radial
-0,6-0,4-0,20,00,20,40,60,81,01,2
0 0,05 0,1 0,15 0,2
No
rmal
ize
d d
isp
lace
me
nt
Radial position (m)
Axial
Radial
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 22 -
Computed electromechanical parameters of the reduced model and stresses per µm
(results of Finite Element Analysis)
Extensional
mode at 26.2
kHz
Flexural
mode at 27.2
kHz
Modal mass M Kg 0.12 0.01
Modal stiffness K N/m 3.18E+09 3.74E+08
Modal electromechanical coupling factor N N/V 2.37 0.33
Shear stress xz /µm at ice/substrate interface Txz/µm MPa/µm 0.11 0.13
Tensile stress xx /µm in PZT Txx/µm MPa/µm 4.55 5.45
Ratio Txx/Txz
- 41.95 42.07
Displacements and voltages required to de-bond the ice
Shear stress required to de-bond ice Txz MPa 1 1
Displacement required to de-bond ice Uo µm 9.22 7.72
Tensile stress in PZT Txx PZT MPa 41.95 42.07
Quality factor Qm - 100 100
Voltage supply U V 124 88
Table 8 – Computation results for the axisymmetric configuration
in the case of bonded PZT+Aluminum Plate+Ice
The next study is the study case (see section 2.1) of this article: a rectangular aluminum plate of dimensions 290x200x1.5
mm3. We consider a square piezoelectric ceramic of dimensions 60x60x2.5 mm
3 bonded in the center of the plate, the
volume of which is similar to that of the bolt-clamped transducer that will be used in the next section. The extensional
radial frequency of the PZT ceramic is around 31.8 kHz. The resonance extensional frequency of the aluminum
plate/ice/PZT assembly close to that frequency is 33.2 kHz. For this extensional frequency, Figure 12 shows the in-plane
and out-of-plane displacements. In-plane magnitudes are 5.6 times higher than out-of-plane magnitudes. The computed
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 23 -
electromechanical force factor is 1.2 N/V. Obtaining sufficient stresses to achieve delamination requires a voltage of
about 86 V and also displacement of 21 µm, which generates a very high stress level in the piezoelectric ceramic.
Palacios emphasizes this issue, which leads to the breakdown of ceramics, in numerous works ([36],[37]).
a) In-plane displacements
b) Out-of-plane displacements
Figure 12 – Displacements for the extensional frequency (33.20 kHz) close to radial frequency of the ceramic
(rectangular configuration with a bonded piezoelectric ceramic)
Computed electromechanical parameters of the reduced model and stresses per µm
(results of Finite Element Analysis)
Extensional mode
at 33.2 kHz
Modal mass M Kg 0.01
Modal stiffness K N/m 4.90E+08
Modal electromechanical coupling factor N N/V 1.2
Shear stress xz /µm at ice/substratum interface Txz/µm MPa/µm 0.05
Tensile stress xx /µm in PZT Txx/µm MPa/µm 3.33
Ratio Txx/Txz
- 70
Displacements and voltages required to de-bond ice
Shear stress required to de-bond ice Txz MPa 1
Displacement required to de-bond ice U0 µm 21
Tensile stress in PZT Txx PZT MPa 70
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 24 -
Quality factor Qm - 100
Voltage supply U V 86
Table 9 – Computation results for the rectangular configuration
in the case of bonded piezoelectric ceramic+aluminum plate+ice
3.3. Piezoelectric de-icing system with Langevin transducers
In order to excite flexural modes of the surfaces to be protected, the architecture of Figure 1(a) is studied here. A bolt-
clamped Langevin transducer (Figure 13) is thus connected to the plate described in section 2.1 through a spacer ring of
length 10 mm. The extensional resonance frequency of the transducer alone is 40.44 kHz. For the Langevin
transducer/aluminum plate/ice assembly, the resonance frequency for which the delamination of the ice occurs with the
lowest supply voltage is 46.857 kHz. For that frequency, the electromechanical force factor is 1.25 N/V. Obtaining
sufficient shear stresses to achieve delamination requires a voltage of about 123 V and the tensile stress in the
piezoelectric ceramics is then 9.8 MPa, which is much less than for architectures with bonded piezoelectric ceramics.
Figure 13 - Bolt-clamped Langevin transducer and its first extensional resonance mode
Figure 14 – Study case: bolt-clamped Langevin transducer connected to the plate
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 25 -
a) Out-of-plane displacements
b) Shear stress at the ice/substrate interface
Figure 15 – Displacements for the flexural mode (46.86 kHz) (rectangular configuration with a Langevin
transducer)
Computed electromechanical parameters of the reduced model and stresses per µm
(results of Finite Element Analysis)
Results at 46.857 kHz
Modal mass M Kg 0.0096
Modal stiffness K N/m 8.4e+09
Modal electromechanical coupling factor N N/V 1.25
Shear stress xz /µm at ice/substrate interface Txz/µm MPa/µm 0.55
Tensile stress xx /µm in PZT Txx/µm MPa/µm 5.3
Ratio Txx/Txz
- 9.63
Displacements and voltages required to de-bond ice
Shear stress required to de-bond ice Txz MPa 1
Displacement required to de-bond ice W0 µm 1.83
Tensile stress in PZT Txx PZT MPa 9.8
Quality factor Qm - 100
Voltage supply U V 123
Table 10 – Computation results for the rectangular configuration
in the case of Langevin transducer+spacer+aluminum plate+ice
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 26 -
3.4. Comparison of the 2 architectures
It is possible to compare the actuating architectures according to several criteria:
The limitations of use owing to stresses in PZT ceramics. The advantage of the architecture with Langevin
transducers is that stresses in the ceramics are only 9.8 MPa, which is 7 times less than for the architecture with
bonded ceramics. Moreover, the risk of mechanical failure is much lower for Langevin transducers than for
bonded ceramics as their pre-stressed structure enables them to withstand higher stresses in operation.
Power consumption is mainly due to mechanical losses during resonance and thus to the mechanical energy of
elastic deformation. This energy is 7.8 times greater for the architecture with bonded ceramics because it
requires much greater deformations (21 microns against 1.83 microns).
The size of the power electronics is dependent on the power consumption but also on the capacitive energy
stored in PZT ceramics. Power devices with good efficiency, such as resonant inverters, have inductors which
are all the larger when the capacitive energy is high. This energy is 2.2 times greater for the architecture with
bonded ceramics because, even if the voltage is lower, the capacitance is greater.
3.5. Experimental validation
Tests were performed on the solution with the Langevin transducer, which seems the most promising if we
consider a criterion that minimizes the energies involved and the stresses in the piezoelectric ceramics. A clamped
aluminum plate without coating was covered with 2 mm of ice (glaze-type ice obtained in a freezing chamber). The
Langevin transducer was then supplied with a voltage of around 46kHz frequency and 150V and 180V amplitude.
Figure 16 shows the results achieved during the experiment. The delamination of the ice started to occur at 150 V
and was even more visible at 180 V. The difference between the estimated voltage and the voltage measured during
the experiment may come from the uncertainty on the quality factor, the non-linearity and the uncertainties on the
shear stress leading to ice delamination (assumed to 1 MPa for computations).
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 27 -
(a) 0V
(b) 150 V
(c)180 V
Figure 16 – Experimental results for the configuration Langevin transducer+spacer+aluminum plate+ice and
excitation of the flexural mode of 46.86 kHz frequency
4. CONCLUSION
This article has aimed to compare different architectures of de-icing systems based on piezoelectric actuators and on
the use of structural resonance modes. The analytical models of the aluminum substrate with ice have shown that it is
more interesting to excite flexural modes than extensional modes to maximize the shear stress at the substrate/ice
interface. These models also highlighted that ultrasonic frequencies over 20 kHz allowed the shear stress to be maximized
for a given displacement, thus avoiding fatigue in the substrate and limiting breakdown of the actuators. The modeling of
the substrate and the actuators by finite element models highlighted the interest of Langevin type actuators with regard to
the resistance to mechanical stress and the mechanical and electrical energies involved in this kind of de-icing system.
Tests validated the efficiency of ice delamination with an architecture based on Langevin type actuators and favoring
structural flexural modes. These experiments also confirm the interest of studying such solutions more intensively.
5. REFERENCES
[1] Civil Aviation Authority. Aircraft Icing Handbook, 2000
[2] Z. Goraj, An overview of the deicing and anti-icing technologies with prospects for the future, 24th Congress of
International Council of the Aeronautical Sciences, 29 August-3 September, Yokohama, Japan, 2004
[3] Federal Aviation Administration. Advisory Circular AC 91-74A - Pilot Guide: Flight in Icing Conditions, 2007
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 28 -
[4] O. Meier, D. Schloz, A handbook method for the estimation of power requirements for electrical de-icing systems,
DLRK, Hamburg, 2010
[5] D. Kevin, L.B. Murphy, Aircraft deicing and anti-icing equipment, Safety Advisor, AOPA Air Safety Advisor
Wheather, Vol.2, pp.1-10, 2004
[6] R. Myose, W. Horn, Y. Hwang, J. Herrero et al., Application of Shape Memory Alloys for Leading Edge Deicing,
SAE Technical Paper 1999-01-1585, 1999, doi:10.4271/1999-01-1585.
[7] S. Ramanathan; V.V. Varadhan, V.K. Varadhan, De-Icing of Helicopter Blades Using Piezoelectric Actuators,
Smart Structures and Materials, Smart Electronics and MEMS, 281 (June 21, 2000), pp. 354-363, 2000
[8] M.K. Kalkowski; T.P. Waters, E. Rustighi, Removing surface accretions with piezo-excited high-frequency
structural waves, Proc. SPIE 9431, Active and Passive Smart Structures and Integrated Systems 2015, San Diego,
California, United States, March 08, 2015
[9] S.V. Venna, Y.J. Lin, In-Flight De-Icing Self-Actuating Wing Structures with Piezoelectric Actuators, Proceedings
of American Society of Mechanical Engineers/International Mechanical Engineering Congress and Exposition 2002,
ASME Press, New York, NY, pp. 237–245., 2002
[10] S.V. Venna, Y.J. Lin, Development of Self-Actuating In-Flight De-Icing Structures with Power Consumption
Considerations, Proceedings of the American Society of Mechanical Engineers International Mechanical Engineering
Congress and Exposition 2003, ASME Press, New York, NY, pp. 45–53, 2003
[11] S.V. Venna, Y.J. Lin, Mechatronic development of self-actuating in-flight deicing systems, IEEE/ASME
Transactions on Mechatronics, vol.11(5), pp.585-592, 2006
[12] S.V. Venna, Y.J. Lin, G. Botura, Piezoelectric Transducer Actuated Leading Edge De-Icing with Simultaneous
Shear and Impulse Forces, Journal of Aircraft, Vol. 44, pp.509-515, 2007
[13] J. Palacios, E. Smith, J. Rose, R. Royer, Instantaneous De-Icing of Freezer Ice via Ultrasonic Actuation, AIAA
journal, vol. 49(6), 1158-1167.
[14] S. Struggl, J. Korak, C. Feyrer, A basic approach for wing leading deicing by smart structures, Sensors and Smart
Structures Technologies for Civil, Mechanical, and Aerospace Systems, Vol. 7981, 2011
[15] R. Seppings, Investigation of Ice Removal From Cooled Metal Surfaces, Mechanical Engineering Department,
Imperial College London, 2006
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 29 -
[16] J.L. Palacios, Design, Fabrication, and Testing of an Ultrasonic De-Icing System for Helicopter Rotor blades, PhD
Thesis, The Pennsylvania State University, Department of Aerospace, May 2008
[17] J.L. Palacios, E.C. Smith, J. Rose, Investigation of An Ultrasonic Ice Protection System For Helicopter Rotor
Blades, Annual Forum Proceedings - AHS International 64th Annual Forum, Vol. 1, pp. 609-618, 2008
[18] J.L. Palacios, E.C. Smith, J. Rose, R. Royer, Ultrasonic De-Icing of Wind Tunnel Impact Icing, Journal of Aircraft,
Vol. 48(3), pp. 1020- 1027, 2011
[19] J.L. Palacios, E.C. Smith, Dynamic Analysis and Experimental Testing of Thin-Walled Structures Driven by Shear
Tube Actuators, AIAA 46th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference,
18 April-21 April, 2005
[20] A. Overmeyer, J. Palacios, E. Smith, R. Royer, Rotating Testing of a Low-Power, Non-Thermal Ultrasonic De-icing
System for Helicopter Rotor Blades, SAE Technical Paper, International Conference on Aircraft and Engine Icing and
Ground Deicing, June 13-17 2011, Chicago, IL 2011
[21] A. Overmeyer, J. Palacios, E. Smith, Actuator Bonding Optimization and System Control of a Rotor Blade
Ultrasonic Deicing System, 53rd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials
Conference, 23 April-26 April, 2012
[22] S. Tarquini, C. Antonini, A. Amirfazli, M. Marengo, J. Palacios, Investigation of ice shedding properties of
superhydrophobic coatings on helicopter blades, Journal of Cold Regions Science and Technology, Vol. 100 , pp. 50–58,
2014
[23] T. Strobl, S. Storm, M. Kolb, J. Haag, M. Hornung, Development of a Hybrid Ice Protection System based on
Nanostructured Hydrophobic Surfaces, ICAS Conference, St. Petersburg, Russia, 7-12 September, 2014
[24] J.J. Petrovic, Mechanical Properties of Ice and Snow, Journal of Material Science, Vol. 38, 2003.
[25] R.J. Scavuzzo, M. L. Chu, Structural Properties of Impact Ices Accreted on Aircraft Structures, NASA Contractor
Report 179580, 1987
[26] HH. G Jellinek, Adhesive Properties of Ice – Part II, 1960
[27] C. Laforte, J. Laforte, Deicing Strains and Stresses of Ice Substrates, Journal of Adhesion Science and Technology
26 (2012) 603-620
[28] D.L Loughborough, E.G. Hass, Reduction of Adhesion of Ice to De-icer Surfaces, Journal of Aeronautical Sciences,
Vol.13(3): pp. 126-134, 1946
Ultrasonic ice protection systems: analytical models for architecture trade-off
- 30 -
[29] S. Timoshenko, Strength of Materials, D. Van Nostrand, 1955
[30] M. Géradin, D.J. Rixen, Mechanical vibrations: theory and application to structural dynamics. John Wiley & Sons,
2014
[31] R.D. Blevins, Formulas for Natural Frequency and Mode Shape, 2001.
[32] J.N. Reddy, Mechanics of Laminated Composite Plates and Shells: Theory and Analysis, 2°ed. 2004
[33] J.N. Reddy, Energy and Variational Methods in Applied Mechanics: With an Introduction to the Finite Element
Method,1984
[34] V. Pommier-Budinger, M.Budinger, Sizing optimization of piezoelectric smart structures with meta-modeling
techniques for dynamic applications, International Journal of Applied Electromagnetics and Mechanics, Vol.46(1),
pp.195-206.
[35] D.A. Berlincourt, H.H.A. Krueger, Behaviour of Piezoelectric Ceramics under Various Environmental and
Operation Conditions of Radiating Sonar Transducers, Technical Publication TP-228, Morgan Electro-Ceramics.
[36] J.T. Soltis, Design and testing of an erosion resistant ultrasonic de-icing system for rotorcraft blades, Thesis in
Aerospace Engineering at Pennsylvania State University, 2013.
[37] A.D. Overmeyer, Actuator bondline optimization and experimental deicing of a rotor blade ultrasonic deicing
system, Thesis in Aerospace Engineering at Pennsylvania State University, 2012.